advanced~3h
Production Postmortem: Diagnosing a Degrading Service
A capstone incident narrative that ties the whole JVM Performance category together: a service that looked perfectly healthy in staging starts degrading under real production load, and the fix comes from working through heap dumps, GC logs, and JFR recordings in exactly the order a real on-call engineer would reach for them.
Learning objectives
- Recognize the symptom pattern of a service that is healthy at low load but degrades specifically under sustained production traffic
- Read a GC log excerpt and identify full-GC thrashing as distinct from ordinary young-generation collection
- Interpret a heap dump's dominator-tree-style output to identify a leak pattern versus ordinary, expected heap growth
- Use a JFR allocation profile to locate a specific hot allocation site responsible for the growth
- Connect a diagnosed root cause (an unbounded cache) to a concrete, minimal code fix and explain why it resolves the full chain of symptoms
This is a Pro chapter
Sign in, then upgrade to Pro or Power to unlock this and the full Core Java Mastery library.
Production Postmortem: Diagnosing a Degrading Service